Active vision approach for optimizing illumination in critical surface inspection by machine vision

This paper describes a new method for the adaptive control of imaging parameters in automized machine vision systems. By these new methodes even in case of critical objects, which show metal and specular reflections or having polished surfaces, the imaging parameters like illumination can be adjusted optimally without any prior knowlegde about the surface characteristics. As a result, an image is generated, which is almost free of irrelevant information in the image. This optimized image only contains the ,,real" egdes actually found on the object's surface and is free of effects resulting from specular reflections or shadows. Surface inspection for scratches, texture analysis or for dimensional measurements becomes much more reliable now.